Subject: Re: [boost] [Review Request] Multiprecision Arithmetic Library
From: John Maddock (boost.regex_at_[hidden])
Date: 2012-04-12 13:15:18
>> The diffs include the removal of some non-needed constants, the trimming
>> max_digits10 and a minor improvement to the multiplication.
> Hmm, that fails a number of tests with error rates > 10^10 - see for
> example test_exp.cpp. Can you investigate?
Never mind - I think I have it - needed to replace uses of
cpp_dec_float_max_digits10 with cpp_dec_float_total_digits10 which does what
it says, where as cpp_dec_float_max_digits10 doesn't anymore. With that
change the tests I've run pass so far.... but they're still running.
Maybe we should rename cpp_dec_float_max_digits10 to something else? BTW,
what's the logic behind the +1 in defining that one? Not that it makes much
difference either which way.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk